A more general illumination model for surface reconstruction

Abstracf For the advantages of non-measurement, quickness, convenience and, most especially, the capability of reconstructing an object surface from just one image, the method of SFS (Shape from Shading) has play an important role in RE (Reverse engineering), object location, pattern recognition and related fields. Recently, many scholars focus; on the problem of how to improve the accuracy of SFS by improving the algorithm of reconstruction. However, rare satisfied results of SFS by improving arithmetic were obtained. In this paper, a new idea is offered to improve the accuracy of SFS by referring to a more general illumination model, the Oren-Nayrtr model, which is available for image rendering field. Meanwhlile, the Oren-Nayar model is improved to satisfy the requirement of iterative approach and the component of specular reflection is added to satisfy the requirement of less rough or even smooth surface. Through taking the place of the traditional illumination model, the Lambert model, by an improved Oren-Nayar model, the accuracy of SFS is improved obviously. In addition, the robust of SFS is improved by transforming the nonlinear factor to linear factor in the illumination model.